Research map / Point clouds / Shape and surface completion
ParSeNet: A Parametric Surface Fitting Network for 3D Point Clouds
Gopal Sharma, Difan Liu, Subhransu Maji, Evangelos Kalogerakis, Siddhartha Chaudhuri, Radomír Měch
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add patch distance loss to control-point loss (ablation) | SplineNet · SplineDataset open B-splines (with upsampling) | Chamfer distance (x100): 2.04 -> 1.96 | Improved |
| add patch distance loss to control-point loss (ablation) | SplineNet · SplineDataset closed B-splines (with upsampling) | Chamfer distance (x100): 5.04 -> 4.9 | Improved |
| add Laplacian loss to cp and dist losses (ablation) | SplineNet · SplineDataset open B-splines (with upsampling) | Chamfer distance (x100): 1.96 -> 1.68 | Improved |
| add Laplacian loss to cp and dist losses (ablation) | SplineNet · SplineDataset closed B-splines (with upsampling) | Chamfer distance (x100): 4.9 -> 3.74 | Improved |
| add post-process optimization to SplineNet (ablation) | SplineNet · SplineDataset open B-splines (with upsampling) | Chamfer distance (x100): 1.68 -> 0.92 | Improved |
| add post-process optimization to SplineNet (ablation) | SplineNet · SplineDataset closed B-splines (with upsampling) | Chamfer distance (x100): 3.74 -> 0.63 | Improved |
| add input upsampling for closed B-splines (ablation) | SplineNet · SplineDataset closed B-splines | Chamfer distance (x100): 3.93 -> 5.04 | Mixed |
| add end-to-end joint training of decomposition and fitting (ablation) | ParSeNet · ABCParts | seg mIoU: 81.20 -> 82.14 | Improved |
| add end-to-end joint training of decomposition and fitting (ablation) | ParSeNet · ABCParts | residual error (all): 0.0120 -> 0.0118 | Improved |
| add post-process optimization after end-to-end training (ablation) | ParSeNet · ABCParts | residual error (all): 0.0118 -> 0.0111 | Improved |
| add point normals to point input (ablation) | ParSeNet · ABCParts | seg mIoU: 71.32 -> 81.20 | Improved |
| use ParSeNet instead of SPFN with points and normals | ParSeNet + e2e · ABCParts | seg mIoU: 69.01 -> 82.14 | Improved |
Closest papers by meaning
- DeepFit: 3D Surface Fitting via Neural Network Weighted Least Squares (ECCV 2020 · similarity 0.90)
- A Survey and Benchmark of Automatic Surface Reconstruction From Point Clouds (TPAMI 2024 · similarity 0.89)
- PUGeo-Net: A Geometry-centric Network for 3D Point Cloud Upsampling (ECCV 2020 · similarity 0.89)
- PIE-NET: Parametric Inference of Point Cloud Edges (NeurIPS 2020 · similarity 0.89)
- Deep Geometric Prior for Surface Reconstruction (CVPR 2019 · similarity 0.88)
- DiffCD: A Symmetric Differentiable Chamfer Distance for Neural Implicit Surface Fitting (ECCV 2024 · similarity 0.88)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
